Research map / Continual learning / Class-incremental
Class Incremental Medical Image Segmentation via Prototype-Guided Calibration and Dual-Aligned Distillation
Shengqian Zhu, Chengrong Yu, Qiang Wang, Ying Song, Guangjun Li, Jiafei Wu, Xiaogang Xu, Zhang Yi, Junjie Hu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add prototype-guided calibrated distillation and dual-aligned prototype distillation to the MiB-style baseline (ablation) | class incremental segmentation (3D CT) · BTCV 4-4 | DSC (%): 89.12 -> 90.60 | Improved |
| add both local and global prototype alignment (DAPD) together instead of one term (ablation) | PGCD+DAPD · BTCV 4-4 | DSC (%): 90.17 vs 89.77 / 90.08 for single terms | Improved |
| add region-based calibrated distillation (ORCD and CRCD) on top of DAPD (ablation) | PGCD+DAPD · BTCV 4-4 | DSC (%): 90.17 -> 90.60 | Improved |
| add prototype-guided calibrated and dual-aligned distillation versus MiB | class incremental segmentation · BTCV 4-4 / 4-2 / 4-1 / 7-1 | DSC (%) all classes: 89.12 -> 90.60 (4-4), 89.94 -> 90.46 (4-2), 90.29 -> 90.55 (4-1), 90.45 -> 90.55 (7-1) | Improved |
| add prototype-guided calibrated and dual-aligned distillation versus MiB | class incremental segmentation · WORD 4-4 / 4-2 / 2-2 / 7-1 | DSC (%) all classes: 82.65 -> 83.42 (4-4), 51.90 -> 83.50 (4-2), 71.14 -> 72.02 (2-2), 72.32 -> 72.89 (7-1) | Improved |
| recompute the background prototype at each incremental step instead of keeping it fixed | DAPD · BTCV / WORD | DSC (%) | No clear effect |
| use the method on 4-4 BTCV instead of training offline on all classes | PGCD+DAPD · BTCV 4-4 | DSC (%) all classes: 90.75 (Offline) vs 90.60 (Ours) | Mixed |
Closest papers by meaning
- Random Anchors with Low-rank Decorrelated Learning: A Minimalist Pipeline for Class-Incremental Medical Image Classification (ICLR 2026 · similarity 0.86)
- Interactive Medical Image Segmentation: A Benchmark Dataset and Baseline (CVPR 2025 · similarity 0.85)
- $\mathbf{D^{3}S^{2}}$: Diffusion-Guided Dataset Distillation for Semantic Segmentation (NeurIPS 2026 · similarity 0.84)
- Incremental Nuclei Segmentation from Histopathological Images via Future-class Awareness and Compatibility-inspired Distillation (CVPR 2024 · similarity 0.84)
- DyCON: Dynamic Uncertainty-aware Consistency and Contrastive Learning for Semi-supervised Medical Image Segmentation (CVPR 2025 · similarity 0.84)
- Why Invariance is Not Enough for Biomedical Domain Generalization and How to Fix It (NeurIPS 2026 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
